Electronics, microelectronics and microelectromechanics require as starting materials semiconductor wafers with extreme requirements made of global and local flatness, single-side-referenced flatness (nanotopology), roughness and cleanness. Semiconductor wafers are wafers composed of semiconductor materials such as elemental semiconductors (silicon, germanium), compound semiconductors (for example composed of an element of the third main group of the periodic table such as aluminum, gallium or indium and an element of the fifth main group of the periodic table such as nitrogen, phosphorus or arsenic) or the compounds thereof (for example Si1-xGex, 0<x<1).
In accordance with the prior art, semiconductor wafers are produced by means of a multiplicity of successive process steps which can generally be classified into the following groups:    (a) producing a usually monocrystalline semiconductor rod;    (b) slicing the rod into individual wafers;    (c) mechanical processing;    (d) chemical processing;    (e) chemomechanical processing;    (f) if appropriate additional production of layer structures.
A method designated “planetary pad grinding (PPG)” is known as a particularly advantageous method from the group of mechanical processing steps. The method is described for example in DE102007013058A1, and an apparatus suitable therefor is described for example in DE19937784A1. PPG is a method for the simultaneous double-side grinding of a plurality of semiconductor wafers, wherein each semiconductor wafer lies such that it is freely movable in a cutout in one of a plurality of carriers (guide cages, “insert carriers”) caused to rotate by means of a rolling apparatus and is thereby moved on a cycloidal trajectory. The semiconductor wafers are processed in material-removing fashion between two rotating working disks. Each working disk comprises a working layer containing bonded abrasive.
The working layers are present in the form of structured abrasive pads which are fixed on the working layers adhesively, magnetically, in a positively locking manner (for example hook and loop fastener) or by means of vacuum. Suitable working layers in the form of easily exchangeable abrasive pads designed to be self-adhesive on the rear side are described for example in U.S. Pat. No. 5,958,794. The abrasive used in the abrasive pads is preferably diamond.
A similar method is so-called “flat honing” or “fine grinding”. In this case, a plurality of semiconductor wafers in the arrangement described above for PPG are guided on the characteristic cycloidal paths between two large rotating working disks. Abrasive grain is fixedly bonded into the working disks, such that the material removal is effected by means of grinding. In the case of flat honing, the abrasive grain can be bonded directly into the surface of the working disk or be present in the form of an areal covering of the working disk by means of a multiplicity of individual abrasive bodies, so-called “pellets”, which are mounted onto the working disk (P. Beyer et al., Industrie Diamanten Rundschau IDR 39 (2005) III, page 202).
Over the course of time, in the case of the grinding methods described, the shape of the working layers changes as a result of constant wear, residual grain breaking out from the bonding matrix and fresh grain being uncovered. It is known that the wear proceeds radially non-uniformly across the working disk. Over time, the working layers in this way form a trough-shaped radial profile, such that the resulting shape of the processed semiconductor wafers worsens to an increasing degree over the course of the wear.
Depending on the materials of the grinding tool and of the processed workpieces, moreover, the cutting capacity of the grinding tool can decrease over time. In addition, it is generally necessary for a new grinding tool to be dressed prior to the first use, by means of the bonding matrix being superficially removed and the abrasive grain embedded therein being uncovered.
Therefore, the prior art includes trimming new or used grinding tools. During trimming, suitable trimming tools are moved under pressure and relative to the tools to be trimmed, such that material removal from the working disks or layers takes place. “Trimming” is understood to mean both the reestablishment of the target shape of the grinding tool (“truing”), and the dressing thereof, i.e. the reestablishment of its cutting capacity.
P. Beyer et al., Industrie Diamanten Rundschau IDR 39 (2005) III, page 202 and DE102006032455A1 describe trimming apparatuses comprising trimming rings and an outer toothing which can be inserted into the grinding apparatus like a carrier and can be moved relative to the working disks by the drives of said grinding apparatus.
The trimming rings described in P. Beyer et al., Industrie Diamanten Rundschau IDR 39 (2005) III, page 202 support a working layer having material-removing action, said working layer containing ceramically bonded diamond as abrasive. Said trimming rings are only suitable for dressing the working layers described by P. Beyer et al., which are composed of a multiplicity of sintered (vitreous), metallically bonded or synthetic-resin-bonded abrasive bodies—so-called pellets. Upon the use of the trimming apparatus described therein and of the method specified therein for trimming abrasive pads, however, the specified trimming rings subject the abrasive pad to wear, without attaining an appreciable dressing effect. Moreover, the dressing apparatus specified therein has proved to be unsuitable for producing a defined target shape of the working layer.
“3M™ Trizact™ Diamond Tile 677XA Pad Conditioning Procedure Rev. A”, 3M Technical Application Bulletin, September 2003, specifies a method for the initial dressing (“break-in”) of a working layer containing bonded abrasive, in which thin, circular abrasive films are stuck to steel disks. The steel disks are toothed and roll on inner and outer pin wheels of the grinding apparatus. Material removal from the working layer is achieved by relative movement between steel disk and working disks under pressure and with addition of water. This method is actually suitable for subsequently dressing working layers that have become blunt, or for providing newly applied working layers, on the surface of which abrasive has not yet been exposed and which therefore do not yet exhibit a cutting effect, with initial dressing for providing a first cutting effect. The method is extremely impracticable, however, since the thin abrasive films applied by adhesive bonding are usually worn within a single use, which results in an extremely unstable dressing process with fluctuating dressing results. Moreover, the abrasive films specified proved to be unsuitable for obtaining trimming of the working layer to form a defined target shape—preferably plane-parallel surfaces of the two working layers.
DE102006032455A1 states that the trimming is advantageously effected predominantly using free grain. The trimming rings disclosed therein continually release abrasive as a result of constant wear, said abrasive ultimately providing for the necessary material removal from the working layers. It has been found, however, that targeted dressing and, in particular, targeted production of a defined target shape of the working layers are not possible using trimming rings of this type.
In addition to the abovementioned specific disadvantages of the above methods, the following problems generally occur during trimming in accordance with the prior art:
The trimming leads to a direction dependence of the grinding behavior of the dressed working layers. It has been observed, for example, that some abrasive pads used as a working layer already have a preferred direction in a manner governed by production. The fashioning of a preferred direction also occurs as a result of the use and as a result of the trimming itself. Preferred direction should be understood here to mean that the abrasive pad, with identical pressure, identical rotational speeds and rotational speed ratios (“kinematics”) of the drives, identical shape of the working gap between the working disks and identical cooling lubrication, achieves a higher material removal rate in one direction than in the case of operation with rotational speeds that are in each case exactly the opposite in terms of sign, but identical rotational speed ratios and identical pressure, gap shape and cooling lubrication. The directional dependence of the grinding behavior has the effect that only very limited rotational speed combinations can be used for the drives of the grinding apparatus.
In addition, during operation in only one direction, the thin carriers for the semiconductor wafers only ever roll in one direction and wear non-uniformly and hence more rapidly by comparison with more uniform loading during operation with a changing direction. This reduces the usable life of the expensive carriers and makes the method uneconomic.
The working layer, too, constantly alters its properties during grinding operation only in one direction. Changing operation with directions of rotation of the drives of the grinding apparatus that alternate from pass to pass or at least from pass block to pass block counteracts that and thus permits more uniform operating conditions.
If the working layers have a preferred direction, however, operation with alternating drive directions is not possible since thickness, shape, removal rate and surface roughness of the workpieces would then constantly alternate, constantly changing heat inputs would make extremely stringent requirements of the regulation of a desirably uniform processing process and, moreover, the working layers would be worn differently and would have to be frequently trimmed or dressed, which necessitates additional process interruptions that adversely affect the economic viability of the method.
These restrictions make the otherwise advantageous PPG method and the measures known in the prior art for keeping constant the shape and cutting behavior of the working layers unsuitable for producing semiconductor wafers of high flatness for particularly demanding applications.
When the working layers are used for grinding workpieces such as semiconductor wafers, for example, the upper and the lower working layers can be subjected to wear of differing magnitudes. The known trimming methods are not able to take account of this different wear, for which reason generally more material than necessary is removed from one of the working layers during trimming. This unnecessary material removal has the effect that the working layers have to be changed more frequently than necessary.
The toothed rings or pin wheels of the rolling apparatus of the grinding machine have a small height coordinated with the thickness of the workpieces usually processed, and are also height-adjustable only to a small extent. Consequently, it is not possible to use trimming bodies of any desired thickness which lead to a height of corresponding magnitude for the trimming apparatuses. This has the effect that the trimming apparatuses or at least the trimming bodies have to be frequently changed.
FIG. 5 shows the essential elements of an apparatus according to the prior art whose working layers can be trimmed by means of the methods according to the invention. The illustration shows the basic schematic diagram of a two-disk machine for processing disk-shaped workpieces such as semiconductor wafers, as is disclosed for example in DE19937784A1, in perspective view. An apparatus of this type has an upper working disk 51 and a lower working disk 52 with collinear rotational axes 53 and with substantially plane-parallel arrangement of the working surfaces of the working disks with respect to one another. According to the prior art, the working disks 51 and 52 are fabricated from gray cast iron, cast stainless steel, ceramic, composite materials or the like. The working surfaces are uncoated or provided a coating made of, for example, stainless steel or ceramic, etc. The upper working disk contains numerous holes 54 through which a cooling lubricant (e.g. water) can be fed to the working gap 55. The apparatus is provided with a rolling apparatus for carriers 56. The rolling apparatus consists of an inner drive ring 57 and outer drive ring 58. The carriers 56 each have at least one cutout which can receive a workpiece 59 to be processed, for example a semiconductor wafer. The rolling apparatus can be embodied for example as pin gearing, as involute gearing or as some other customary type of gearing. Upper working disk 51 and lower working disk 52 and inner drive ring 57 and outer drive ring 58 are driven at rotational speeds no, nu, ni and na about substantially identical axes 53. In this case, “substantially” means that the offset of the axes of rotation of the individual drives relative to the central axis of all the drives amount to less than one per mille of the diameter of the working disks, and the tilting of the axes with respect to one another amounts to less than 2°. A cardanic suspension of the upper working disk 51 compensates for any residual tilting of the axes, such that the mutually facing working surfaces of the working disks can be moved with azimuthally identically distributed force and without wobbling movement relative to one another.
Each working disk 51, 52 supports a working layer 60, 61 on its working surface. The working layers are preferably abrasive pads.
An “abrasive pad” is understood hereinafter to mean a working layer composed of at least three layers, comprising                a closed, continuous or interrupted useful layer, facing away from the working disk, in the form of a smooth or structured film, a woven fabric, felt, knitted fabric or individual elements, which contains bonded abrasive and has a useful thickness of more than one abrasive grain layer and at least one part of which makes direct contact with the workpieces to be processed and thereby brings about material removal;        a central closed, or at least continuous support layer in the form of a smooth or structured film, a woven fabric, knitted fabric or felt, which supports the useful layer and connects all the elements of the useful layer to form a continuous unit; and        a closed, continuous or interrupted mounting layer, which faces the working disk and, over the period of the useable life of the useful layer or a shorter period, determined by the user, forms a force-locking or positively locking composite assembly with the working disk of the grinding apparatus, for example by means of vacuum (sealed mounting layer), magnetically (mounting layer contains a ferromagnetic layer), hook and loop fastener (mounting layer and working disk contain “hook” and “loop”), adhesive bonding (mounting layer is provided with self-adhesive or activatable adhesive layer), etc. The abrasive pad is elastic and can be detached from the working disk by peeling movement. The abrasive pad can, particularly when covering particularly large working disks, be subdivided into up to eight segments, four segments for each working platen, which can be removed or mounted individually to form a gap-free parquetting of the working disk area to be covered.        
Suitable abrasive pads are described for example in U.S. Pat. No. 5,958,794. The abrasive pads are preferably structured in the form of small regular units. Preferably, these units consist of regularly arranged “islands” (uniformly elevated regions) and “trenches” (recessed regions). In this case, the islands become engaged with the workpieces and thus bring about material removal. The trenches feed in cooling lubricant and carry away resulting grinding slurry. The absolute size of islands and trenches and the area ratio thereof (supporting area proportion of the working layer) constitute crucial features for the material-removing function of the working layer. The islands of one abrasive pad that is preferably used (Trizact™ Diamond Tile 677XA or 677XAEL from 3M Company) have for example a square shape having an edge length of a few millimeters and are separated by trenches having a width of approximately one millimeter, thus resulting in a supporting area proportion of between 50% and 60%.
The abrasive used in the abrasive pads is preferably diamond. However, other hard substances are likewise suitable (for example cubic boron nitride (CBN), boron carbide (B4C), silicon carbide (SiC, “carborundum”), aluminum oxide (Al2O3, “corundum”), zirconium oxide (ZrO2), silicon dioxide (SiO2, “quartz”), cerium oxide (CeO2) and many others.
However, the abrasive grain can also be directly bonded into the surface of the working disk or be present in the form of an areal covering of the working disk by means of a multiplicity of individual grinding bodies, so-called “pellets”, which are mounted onto the working disk.
The working gap formed between the working layers 60 and 61 fixed on the upper working disk 51 and lower working disk 52, within which gap the semiconductor wafers are processed, is designated by 55 in FIG. 1.